Commenced in January 2007
Frequency: Monthly
Edition: International
Paper Count: 8

Quantization Related Publications

8 Fast Adjustable Threshold for Uniform Neural Network Quantization

Authors: Alexander Goncharenko, Andrey Denisov, Sergey Alyamkin, Evgeny Terentev

Abstract:

The neural network quantization is highly desired procedure to perform before running neural networks on mobile devices. Quantization without fine-tuning leads to accuracy drop of the model, whereas commonly used training with quantization is done on the full set of the labeled data and therefore is both time- and resource-consuming. Real life applications require simplification and acceleration of quantization procedure that will maintain accuracy of full-precision neural network, especially for modern mobile neural network architectures like Mobilenet-v1, MobileNet-v2 and MNAS. Here we present a method to significantly optimize training with quantization procedure by introducing the trained scale factors for discretization thresholds that are separate for each filter. Using the proposed technique, we quantize the modern mobile architectures of neural networks with the set of train data of only ∼ 10% of the total ImageNet 2012 sample. Such reduction of train dataset size and small number of trainable parameters allow to fine-tune the network for several hours while maintaining the high accuracy of quantized model (accuracy drop was less than 0.5%). Ready-for-use models and code are available in the GitHub repository.

Keywords: Neural Networks, Machine Learning, Distillation, Quantization

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 232
7 Optimization Approach to Estimate Hammerstein–Wiener Nonlinear Blocks in Presence of Noise and Disturbance

Authors: Leili Esmaeilani, Jafar Ghaisari, Mohsen Ahmadian

Abstract:

Hammerstein–Wiener model is a block-oriented model where a linear dynamic system is surrounded by two static nonlinearities at its input and output and could be used to model various processes. This paper contains an optimization approach method for analysing the problem of Hammerstein–Wiener systems identification. The method relies on reformulate the identification problem; solve it as constraint quadratic problem and analysing its solutions. During the formulation of the problem, effects of adding noise to both input and output signals of nonlinear blocks and disturbance to linear block, in the emerged equations are discussed. Additionally, the possible parametric form of matrix operations to reduce the equation size is presented. To analyse the possible solutions to the mentioned system of equations, a method to reduce the difference between the number of equations and number of unknown variables by formulate and importing existing knowledge about nonlinear functions is presented. Obtained equations are applied to an instance H–W system to validate the results and illustrate the proposed method.

Keywords: Optimization, Identification, Quantization, Hammerstein-Wiener

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 417
6 New Efficient Method for Coding Color Images

Authors: Wajeb Gharibi, Walaa M.Abd-Elhafiez

Abstract:

In this paper a novel color image compression technique for efficient storage and delivery of data is proposed. The proposed compression technique started by RGB to YCbCr color transformation process. Secondly, the canny edge detection method is used to classify the blocks into the edge and non-edge blocks. Each color component Y, Cb, and Cr compressed by discrete cosine transform (DCT) process, quantizing and coding step by step using adaptive arithmetic coding. Our technique is concerned with the compression ratio, bits per pixel and peak signal to noise ratio, and produce better results than JPEG and more recent published schemes (like CBDCT-CABS and MHC). The provided experimental results illustrate the proposed technique that is efficient and feasible in terms of compression ratio, bits per pixel and peak signal to noise ratio.

Keywords: Image Compression, Quantization, color image, q-coder, edge-detection

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1282
5 Coding of DWT Coefficients using Run-length Coding and Huffman Coding for the Purpose of Color Image Compression

Authors: VINOD KUMAR, Varun Setia

Abstract:

In present paper we proposed a simple and effective method to compress an image. Here we found success in size reduction of an image without much compromising with it-s quality. Here we used Haar Wavelet Transform to transform our original image and after quantization and thresholding of DWT coefficients Run length coding and Huffman coding schemes have been used to encode the image. DWT is base for quite populate JPEG 2000 technique.

Keywords: Quantization, DWT, Run Length Coding, huffman coding, JPEG2000, lossy compression

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2425
4 Optimization of Quantization in Higher Order Modulations for LDPC-Coded Systems

Authors: M.Sushanth Babu, P.Krishna, U.Venu, M.Ranjith

Abstract:

In this paper, we evaluate the choice of suitable quantization characteristics for both the decoder messages and the received samples in Low Density Parity Check (LDPC) coded systems using M-QAM (Quadrature Amplitude Modulation) schemes. The analysis involves the demapper block that provides initial likelihood values for the decoder, by relating its quantization strategy of the decoder. A mapping strategy refers to the grouping of bits within a codeword, where each m-bit group is used to select a 2m-ary signal in accordance with the signal labels. Further we evaluate the system with mapping strategies like Consecutive-Bit (CB) and Bit-Reliability (BR). A new demapper version, based on approximate expressions, is also presented to yield a low complexity hardware implementation.

Keywords: Mapping, Quantization, Low Density parity Check, Demapping, Quadrature Amplitude Modulation

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1421
3 Binary Phase-Only Filter Watermarking with Quantized Embedding

Authors: Liu Yi, Hu Haibo, He Ming

Abstract:

The binary phase-only filter digital watermarking embeds the phase information of the discrete Fourier transform of the image into the corresponding magnitudes for better image authentication. The paper proposed an approach of how to implement watermark embedding by quantizing the magnitude, with discussing how to regulate the quantization steps based on the frequencies of the magnitude coefficients of the embedded watermark, and how to embed the watermark at low frequency quantization. The theoretical analysis and simulation results show that algorithm flexibility, security, watermark imperceptibility and detection performance of the binary phase-only filter digital watermarking can be effectively improved with quantization based watermark embedding, and the robustness against JPEG compression will also be increased to some extent.

Keywords: Digital Watermarking, Quantization, image authentication, Discrete Fourier transform, binary phase-only filter

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1207
2 Design of Multiplier-free State-Space Digital Filters

Authors: Tamal Bose, Zhurun Zhang, Miloje Radenkovic, Ojas Chauhan

Abstract:

In this paper, a novel approach is presented for designing multiplier-free state-space digital filters. The multiplier-free design is obtained by finding power-of-2 coefficients and also quantizing the state variables to power-of-2 numbers. Expressions for the noise variance are derived for the quantized state vector and the output of the filter. A “structuretransformation matrix" is incorporated in these expressions. It is shown that quantization effects can be minimized by properly designing the structure-transformation matrix. Simulation results are very promising and illustrate the design algorithm.

Keywords: Digital Filters, Quantization, state-space, minimum noise, multiplier-free

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1167
1 Performance Analysis of Chrominance Red and Chrominance Blue in JPEG

Authors: Mamta Garg

Abstract:

While compressing text files is useful, compressing still image files is almost a necessity. A typical image takes up much more storage than a typical text message and without compression images would be extremely clumsy to store and distribute. The amount of information required to store pictures on modern computers is quite large in relation to the amount of bandwidth commonly available to transmit them over the Internet and applications. Image compression addresses the problem of reducing the amount of data required to represent a digital image. Performance of any image compression method can be evaluated by measuring the root-mean-square-error & peak signal to noise ratio. The method of image compression that will be analyzed in this paper is based on the lossy JPEG image compression technique, the most popular compression technique for color images. JPEG compression is able to greatly reduce file size with minimal image degradation by throwing away the least “important" information. In JPEG, both color components are downsampled simultaneously, but in this paper we will compare the results when the compression is done by downsampling the single chroma part. In this paper we will demonstrate more compression ratio is achieved when the chrominance blue is downsampled as compared to downsampling the chrominance red in JPEG compression. But the peak signal to noise ratio is more when the chrominance red is downsampled as compared to downsampling the chrominance blue in JPEG compression. In particular we will use the hats.jpg as a demonstration of JPEG compression using low pass filter and demonstrate that the image is compressed with barely any visual differences with both methods.

Keywords: Image Compression, Quantization, JPEG, discrete cosine transform, Color Space Conversion, Peak Signal to Noise Ratio & Compression Ratio

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1392